Converting Physical to Virtual - What about empty space?

[Warwick_Tech]

Hi All,

 

So we are in the process of converting our last few Physical devices to VHD/VHDX

 

We're using a mix of methods, from Disk2VHD to Windows Server Backup (Restore the VHDX and then create a new Generation2 machine and mount it)

 

But, our main issue is our old file server is over 20TB - of which only a few TB are now in use (Isn't the cloud wonderful)

 

Long story short, I've got a smaller Server for it to live, but I can't find anything definitive on disk space and calculations -

 

Does Disk2VHD or other Physical > Virtual services render the 'empty' data as HDD size? I have shrunk the partition down to a few TB so the rest is 'unallocated' but will this work? What I can't have is a 20TB VHDX file as that won't fit on my new system.

 

The backups say it's only the few TB, but honestly I can't find anything concrete on this.

 

Someone suggested if there was a way to shrink VHDX files, but this will only shrink the file, not the drive size inside (?)

 

Anyone got anything definitive?

[3s-gtech]

If you mount the Disk2VHD disk as 'fixed size' it will be the full size of the original image. If you use a dynamic disk it will shrink back down - but the server will be slow.

 

You may need to defrag the disks (with a proper defrag that will compact the space used), shrink the partitions on the host, then Disk2VHD. You shouldn't then carry across the unallocated space.

[Warwick_Tech]

If you mount the Disk2VHD disk as 'fixed size' it will be the full size of the original image. If you use a dynamic disk it will shrink back down - but the server will be slow.

 

You may need to defrag the disks (with a proper defrag that will compact the space used), shrink the partitions on the host, then Disk2VHD. You shouldn't then carry across the unallocated space.

 

Good ideas, the issue is I can't load the 20TB file onto the new server as it's only 10TB (grr)

 

I have done a defrag and cleared as much as possible, I suppose in theory Disk2VHD will only create a disk that size as it uses mapped drive letters (C: D: etc) and the unallocated space has no drive letter....

 

Yes, we may have cracked it... maybe

[Warwick_Tech]

I don't suppose anyone knows when using Disk2VHD if I have several drives, do I need to run it once for each drive, or run it all at once?

 

The way I have my VM's ATM is one drive is the OS, and the second drive contains all the data.

[Steve21]

As mentioned above When I last did this I just shrunk the disk volume on the original server before running disk2vhd

 

As it copies the volume rather than the VHD behind it if you shrink the 20TB volume to say space+1TB, then copy the image, import new server and then expand the volume to 10TB etc

 

Again As above you might need to defrag etc so all the data sits at the start but that’s my normal route

 

One thing to note, if it won’t shrink enough it’s likely to be the virtual memory cache etc, you can turn that off restart vm and it’s likely it’ll let you shrink it further

 

Regarding the drives query, it’ll create a VHD per “disk” normally, not volume as it’ll copy partition info etc so depends how you have it split now

 

Steve

[KK20]

I backed up in veeam and restored to smaller destination, it worked well in my P2V. Depending on how many P2V you are doing, community veeam is free to 10 licenses, I have no idea if the free agent version will work on servers, if so then that will simplify matters enormously. The beauty of this is that you dont need to mess with the source disk (other than defrag and installing a veeam agent). Granted this means you will need a temporary backup destination for veeam repository but you can use a NAS or even a giant USB drive.

 

unmovable blocks will be your only barrier, you know if you have them as you can start the "shrink" process after defrag to see what windows could shrink to (you dont have to commit). This will be the smallest VHDX drive you could use in this method.